Flow Forecast is an open source deep learning for multivariate time series forecasting, classification, and anomaly detection platform written in PyTorch. Our goal is to build a general open source time series framework (somewhat analogous to HuggingFace has done for NLP) and enable users to solve complex real world forecasting problems. Additionally, we aim to solve pressing AI4Good problems. For instance, we are currently using Flow-Forecast to create county level COVID-19 forecasts as well as project flash floods around the U.S.
We are currently, looking for new contributors to add new models, benchmark existing models, correct bugs, and enhance our cloud provider integrations. We welcome developers and data scientists of all skill levels to come and contribute. We wall also be hosting an official Sprint at PyData Global.
- You can look at some of the things we are working on here
Hope to see you on the project!